When Do Campaigns Matter? Informed Votes, the Heteroscedastic Logit and the Responsiveness of Electoral Outcomes

Abstract

Previous research suggests that voters in mass elections tend to be badly informed. If these voters do not know enough about the relationship between the policy consequences of electoral outcomes and their own interests, then electoral outcomes may not provide meaningful expressions of voter interests. Can campaign activity affect the relationship between voter interests and electoral outcomes? To answer this question, we use survey data from 35 comparable elections and a new empirical methodology (Dubin and Zeng's [1991] heteroscedastic logit). The new methodology allows us to estimate the joint effect of voter information and interests on voting behavior in a way that is both theoretically justifiable and better at explaining the available data than traditional methods. We find that campaign activity increases the likelihood that electoral outcomes are responsive to (perhaps, otherwise badly informed) voter interests, when campaigners are able to exert costly and observable effort, are able to make credible statements and have the opportunity to engage in a vigorous and competitive campaign.

Files

Additional details